5.0.23 Upgrade to 6.0.8 (Testing in vmware first)

[hodd]

So we're planning to roll up our 5.0.23 server to 6.0.8 on Wednesday night, and I wanted to just test out the update using a P2V copy of our physical mail server. I assigned it a DHCP address, and fired it up. After running the installer on the virtual machine, when it gets to the verification of the database it gives me this error:

/opt/zimbra/mysql/bin/mysqlcheck: Got error: 2002: Can't connect to local MySQL server through socket '/opt/zimbra/db/mysql.sock' (2) when trying to connect
No errors found
command failed

I can carry on, but eventually it will die here:

Starting mysql...done.
UPGRADE FAILED - exiting.


Is this a bad way to test the upgrade of my server?

[LMStone]

When you launched the Zimbra 6.0.8 installer, was Zimbra still running?

If not, leave your existing Zimbra installation running while launching the installer and see if that helps.

We have seen a bit of what appears to be a timing bug sometimes in the installer at that same juncture: the installer seems not to wait long enough for MySQL to start if it is not yet running and the installer then fails. Doesn't happen often, causes no harm (just rerun the installer with Zimbra running) and haven't been able to reproduce it reliably, so never filed a bug report.

Hope that helps,
Mark

[tonster]

Originally Posted by hodd

So we're planning to roll up our 5.0.23 server to 6.0.8 on Wednesday night, and I wanted to just test out the update using a P2V copy of our physical mail server. I assigned it a DHCP address, and fired it up. After running the installer on the virtual machine, when it gets to the verification of the database it gives me this error:

/opt/zimbra/mysql/bin/mysqlcheck: Got error: 2002: Can't connect to local MySQL server through socket '/opt/zimbra/db/mysql.sock' (2) when trying to connect
No errors found
command failed

I can carry on, but eventually it will die here:

Starting mysql...done.
UPGRADE FAILED - exiting.


Is this a bad way to test the upgrade of my server?

You want to make sure that ZCS is stopped when you make your copy. I'm not sure exactly how you're performing this copy, we generally recommend rsync to sync all the data, then shut down ZCS and perform a final sync (it wont take long) and then you can start it on the backup server. If you're using a different IP, you'll want to probably setup a fake dns server on that box and set the hostname the same or at least add the hostname to /etc/hosts and start the server. Make sure it cannot talk to the live server to prevent accidentally manipulating live data as well.